Professional Documents
Culture Documents
Journal, Vol. 92, No. 2 (Summer, 2008), pp. 166-178 Published by: Wiley on behalf of the National Federation of Modern Language Teachers Associations Stable URL: http://www.jstor.org/stable/25173021 . Accessed: 17/01/2013 00:05
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp
.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.
Wiley and National Federation of Modern Language Teachers Associations are collaborating with JSTOR to digitize, preserve and extend access to The Modern Language Journal.
http://www.jstor.org
This content downloaded on Thu, 17 Jan 2013 00:05:11 AM All use subject to JSTOR Terms and Conditions
Introduction: Development
KEES DE BOT
University Faculty P.O. Box of Groningen ofArts 716
some of the basic characteristics of complex In this contribution, systems, collectively adaptive de are discussed. Such labeled (DST), systems are self-organizing, Dynamic Systems Theory The and show emergent on initial conditions, sometimes chaotic, they properties. pendent over time. is seen as a dynamic is on development focus in DST system, and language Language of as a A number of examples and attrition, both acquisition process. dynamic development, are mentioned. After a short in the field of applied of DST linguistics applications possible are discussed. some lines of research of each of the individual articles, possible presentation
THIS SPECIAL
sentations pects at of Language
to to find a venue a for the papers idea good The Modern When available. them publicly make an interest in showed (MLJ) Journal Language a issue on this topic,1 all con special publishing for such a pub their support tributors expressed lication stone plexity that would, in the on the one introduction set a mile hand, of new ideas on com
and dynamic systems in the field of applied and, on the other hand, allow the larger to get a state-of-the-art I will of overview of these
In this introduction, the development and dynamic complexity, tory of their role in the social
on unpre symposium AAAL-sponsored Di in language by teaching, organized dictability an ane Larsen-Freeman, attracted unexpectedly for a small com It is not unusual large crowd. a new to get excited about of researchers munity interest here the in their field, but ap approach to a larger group that to have spread peared was that something sensed happening. special like it seemed in these events, the interest Given
present theories
a short his
as a dynamic system, and language develop guage to what ment as a dynamic process, we can explore and dynamics extent and how ideas on complexity can be fruitful for our field. The main character briefly more istics of dynamic several because extensive aspect, further systems will be presented of the articles will present
0026-7902/08/166-178 $1.50/0
The Modern Language
Language
One of different elaborations aspects. in criticality, will be discussed self-organized since it has not figured prominently detail on the of Dynamic in the discussion application a short introduction (DST). Finally, Theory Systems a look to each of the articles will be followed by and some thoughts further. on how these ideas may
ahead
be developed
This content downloaded on Thu, 17 Jan 2013 00:05:11 AM All use subject to JSTOR Terms and Conditions
KeesDeBot A SHORT HISTORY OF CHAOS THEORY, COMPLEXITY THEORY, AND DYNAMIC SYSTEMS THEORY
As with most theories, it is rather difficult to the 1990s the Santa to Fe
167
systems, with study complex adaptive a role and Institute playing leading and No started for the
of this particular beginnings commentators Some (Schueler mention 1963 of the publication article on weather in
like John Holland top scientists attracting DST Gell-Mann. bel Prize winner Murray out as a mathematical approach purely of complex development more it is now considered proaches change. than a fixed
Schueler,
systems over time, but a set of tools and ap and all-inclusive theory of
in initial con systems, small differences complex differ ditions can lead to larger and unpredictable toWolfram ences over time. However, according to initial conditions has been sensitivity ever since the work a role in to physics play in the mid-19th of Maxwell century. (2002), shown The term here meaning to the fact that the outcomes of interac chaos has technical over time cannot lack mathematics. predicted chaos Thus, of order. As the be be is a a
are over systems systems that change Dynamic use the follow and Port time. Van Gelder (1995) we take dynam ing definition: "Roughly speaking, states to be systems with numerical ical systems over to some time according rule" that evolve (p. 5). In DST, role. sition geometrical have Systems variables. interacting an concepts important play a state is a po which phase, in a multidimensional that defines space Systems in systems are rep Changes state space. DST in the trajectories to de models and simulations computer state. are sets of
conventional
rather than unpredictability to initial result of sensitivity havior be systems even random, in the no of that exhibit though sense that
employs of systems over time. Change scribe changes typ not continuous: tend to settle in Systems ically is what as "a are called set "attractors," which can be denned in the that has a neigh space phase in which and every point stays nearby as time goes to infin the attractor 2008).
it is well
within
out emerge in initial differences is universal like researchers sentatives refers (1998) of refers order in turbulence
two main The parameters. can are chaos that Chaos Theory of variables and of the interaction random conditions, and that there in seemingly chaotic patterns, of the leading in liquids. Two can be seen as repre (1997) Holland Mandel two trends: of
systems tend to show Dynamic refers to a in which nonlinearity development, between and effects. Although discrepancy input the different theories have their own histories and foci, in many publications the labels Chaos, Systems, Nonlin Adaptive in almost often are used
these
to "the
with
to refer to a class of theories. What terchangeably in common is a focus on the develop they have ment I will of complex systems over time. Here use to refer to this whole set of the term DST notions. Different applied nomics, recently, will focus aspects in a wide of DST have of been studied such as and eco
system. at what "at the happens Complexity Theory looks on chaos (Lewin, 1999). Research edge of chaos" more had shown that complex systems develop or
time and then more for some less predictably to show chaotic behavior. less suddenly begin as "a chaos of defines (1992) complexity Waldrop in which behaviors the components of the system never never lock into yet quite place, quite dissolve or either" (p. 293). is part of the study of systems, in which sys tems are studied as a whole rather than with focus turbulence DST on their parts. from the early ranging adaptive chotherapy. in Systems Theory, which emerged in many fields 1950s, has been applied ontology One branch Several to management psy is the study of complex set up in institutes were and into
on the in the field of of DST application and then further narrow my discussion cognition to the study of language development.
range and, more meteorology, oceanography, I science. In the next section, cognitive
fields,
A DYNAMICAL APPROACH
Several cation most publications of DST notions have
TO COGNITION
focused on the appli The
in human
systems.
Tim supporter outspoken van Gelder, a DST to who has defended approach a number in of influential cognition publications. as Motion: The first is the edited volume Mind
This content downloaded on Thu, 17 Jan 2013 00:05:11 AM All use subject to JSTOR Terms and Conditions
168 TheModern Language Journal 92 (2008) Explorations in the Dynamics of Cognition (1995),
which he edited with Robert Port. The 1998 other article publication "The is van Gelder's but they can be Traditional lin separately. ear methods of linking behavior with underlying can be mental in this processes applied approach. In the second ar van Orden (2002) approach, "nonlinear interactions gues, may occur among be looked and isolated studied processes, cognitive their environments, agents ses will Orden agents and cognitive even among cognitive themselves. If so, then empirical analy nonlinear methods" require (p. 2). Van and between can at as nonlinear,
keynote
which
he claims that "Dynamics is ex used and powerful widely in science" framework (1998, p. 622). planatory commentators of the BBS Several take issue with arguably in cogni that, as yet, DST to provide explanations; at best it can describe but not ex development, it A characteristic of the Gelder, (van 1998). plain this statement tive science and cannot argue claim system cognitive Gelder mentions as a and system dynamical most commentators that van seem on
positions namical
the second and clearly favors approach, in another article on self-organization of cogni van Orden, tive performance, and Tur Holden, (2003) a nonlinear propose dynamic as an alternative for mainstream This led to a strong systems cogni reaction by Far Wagenmakers,
vey
researchers,
to agree with is that cognition three levels: the nervous system, environment. This view also has
is embedded
such as Beer and van (2000) by other researchers, a as Geert The view of (1998). cognition dynami a move cal system represents away from a strictly in which modular is viewed as approach cognition a separate module to the brain confined working in isolation. The debate over whether DST can be seen as an alternative sically tinues for existing which paradigms, van Gelder the position is defending, considerable Beer with a controversy (2000) argues community. and symbolic which play is ba con
that the more argue as as is to be preferred approach long the superiority of the alternative is not proven: "If (2005), science of nonlinear is truly superior systems dynamical to the current of paradigm then its proponents should compar
including who
cognitive psychology, a head-on not shy away from quantitative ison" 108). (p. What from the literature on emerges is that of DST not there an alternative.
to generate
are their claims, and there is overstating to abandon substantial evidence yet enough traditional in favor of a DST science cognitive approach.
based
connectionist critical
a lot in common
approaches
notions: "... of regardless as the dynamical approach an alternative or an to more traditional adjunct ideas are forcing a much approaches, dynamical of the notions needed critical evaluation of repre the eventual of existing fate of sentation and computation in cognitive science"
sideration
in lin existing cognitive have taken on guistics and language development com the theory as an interesting but basically in the of plementary study approach language processing based and
language
development.
Cooper
that a combination
of
and Goldstein
articulatory phonology are
is needed for an understanding of fundamental and mechanisms of change. Van Or processes den (2002) says that "twentieth century cognitive was built on the of linear assumption psychology an for and he that 1), argues ity" (p. approach In his view, there are this assumption. challenges to two when trying apply DST princi approaches to like word processes recognition cognitive ples or reading. In the first approach, subprocesses
movements
to show how
extensive
language
processing
application be found
(1995, 2004);
some of his
This content downloaded on Thu, 17 Jan 2013 00:05:11 AM All use subject to JSTOR Terms and Conditions
KeesDeBot
from traditional different that are markedly to here. will be discussed approaches language on Elman's 1995 article, a number of char Based ideas acteristics of such traditional approaches are non controversial: on discrete 1. Processing is based and context of free symbols, which means that the meaning use. of is their symbols independent rules are operators, while lexi 2. Grammatical cal elements on are lexical operate 3. Representations nature. Words dictionary independent 4. Building operands; items. (words, lexicon remain in other words, rules pect
169
of the system which the system state in cer push tain directions rather than others" (p. 195). This is rather from that of tra different perspective ditional switch field. and and a linguistics psycholinguistics, to this view will lead to a shift in the major
are static in rules) are like words in a the same over time,
that will
words, with the words as the mortar. The explain processing most why important we need is that the
as
"We processes: with the conviction development tion of mental life is continuous cal growth also not applies seen as of form to and
that develop (1994) argue are governed by similar of human the mystery approach with that the acquisi all biologi
a new
Elman provides on theory language traditional view of mental with are no how we now
reason
to
language
is at variance There
think pas
static, discrete,
This (p. XIII). which is development, in the cognitive sys special system that develops
function"
and maybe representations, of representations is wrong. a view on as He proposes processing language a in which the time dimen process dynamic are not static sion is crucial: elements Language use and are context change through highly sensitive. Time is seen not as the moving of the of a clock but as effects hands its representing on so time is part of the inherently processing, but rather than some external dimension. processing To show how language take development might a Elman (1995) place using dynamic approach, to trained a simple recurrent network sub predict a small set of 29 in a sentence sequent words using verbs and nouns with a corpus of 10,000 short sen tences. The that the network simulation showed an internal structure that reflects differ develops ent of word classes. and meanings approximations In other words, can induce the network lexical cat egories learn from statistical Elman simulations, tences. In subsequent regularities. can also showed that networks
as a First (LI) dynamic language acquisition in a has been studied process variety of ways, but a combination the dominant has been approach of microgenetic studies and computer modeling. In microgenetic studies 8c Lowie, (see Verspoor, van this issue), Dijk, fine-grained analyses of dense a rich set of informa data provide acquisitional tion that reveals all the small steps that are taken in development. ten combined show how development. are fruitful ways of simulations of language devel understanding most we The is that need opment. important point to turn our intuitions and what we know about Modeling and enhancing our into the strict language language development For to be very mathematics. that, we need of ex Child with are of data production of caretaker talk to analyses to and type of input are related
amount
sen in long-distance dependencies complex This research that a very simple it suggests can lead to a erative learning mechanism dynamic structure tributed able and of language in which Van words Gelder have a dis representation in a network that is adapt and Port "Thus inter
we think we know because it plicit about what to translate vague is nearly statements impossible into equations. We may be convinced that motiva an tion in role Second plays important Language but it is far from easy to turn (SLD), Development that idea exactly into a mathematical and when how that states equation that factor plays a role. A we need model to ask is why closest test our over answer is that this on the de hypotheses time. Language
The
symbols but or the lexicon space, dictionary is the structure in this space, and rules processing are not but the dynamics symbolic specifications
is seen as an iterative process in which velopment new to is added the and input existing knowledge to the expectations with respect are development
This content downloaded on Thu, 17 Jan 2013 00:05:11 AM All use subject to JSTOR Terms and Conditions
more
as the description of language development the work by van Geert, who has process, dynamic been the field for more than a decade, is leading a taken here of his cussed ideas as a on starting point. The basic elements will be dis language development detail here because are as relevant
place. For
in some
issues he raises
The connection be growers. tween the growers be parallel and simul may a different to have but they are taneous, likely can time path. An of connected growers example be found in a and Mervis study by Robinson at looked and vocabulary growth of plural markers. In order to combine two these and curves, Robinson developmental as Mervis used the "precursor model" proposed van Geert In a precursor there model, (1995). by are two variables, a a successor. and predecessor in the successor Growth is initially suppressed (1998), the use who by growth in until petition, the predecessor. growth possible in both a the predecessor, a threshold level After variables shares form of com in is reached is reached,
as connected
van Geert contributions has made is to re major in terms of formulate development "growth." The are that it ismath of using this concept advantages well defined ematically form it can be modeled function, functions the that from these and which is one and that a in its simplest using logistic growth of the most widely used starting point, functions complex and findings He uses over time to test the a systems. Taking
in the study of dynamic as function logistic growth take has developed into account on more
this threshold
van Geert
assumptions
relations
defined
development.
dense
support. In addition
the data. are that growth that can grow, what structural
to the of the individual modeling van Geert of children, also has ap interaction interaction between is seen as
is de (p. 314), and that growth resources. In pendent language development as well as moti input and output, encouragement, and time to learn, can vation, attention, feedback, growth on be viewed ited and capacity, ited, but as resources. interlinked input, time, can be Resources in a dynamic system. Memory are all lim and motivation typically are lim
(1995) condition"
"minimal
ecosys part important cognitive tem: "A child's does however, ecosystem, cognitive not stop at the boundaries or the of the brain all aspects of the environment, body, but contains as well, insofar as that environment is accessible to and Van understood refers Geert by the child" to co-construction (1995, p. 332). as a dynami He exchange. one for the the fine for over
of the child's
in the sense compensatory they that a lack of time can, to a certain extent, be com for with or more ef higher motivation means interconnectedness of resources interact on over time: Motivation an effect external
of mutual cal process information describes how two sets of equations, in the child and the other growth tuning of the caretakers, on the child developing the caretaker not and are linked the basis
time, with
the caretaker
the resources
resources role in dif will play a different Specific In the early stages ferent stages of development. of L2 development, for example, memory capacity than analytic ability, while may be more important in later terlinked stages
to the experience of success, so on. Also, more to and time, willingness spend comes the level of development into play here:
level of development of the child. Van Geert does claim that the type of modeling he proposes can the described: phenomena explain
There
is now reasonable evidence that in spite of their can processes underlying complexity, developmental be viewed as instances of relatively simple growth mod els. It remains the organization
to be explained, of course, what it is in of the brain or of human culture that processes (p. 335) follow their par
in the opposite may be true. The resources of forms what called is system
makes
This content downloaded on Thu, 17 Jan 2013 00:05:11 AM All use subject to JSTOR Terms and Conditions
171
self-organizing to open is by system, which is Learning inspection. van iterations as discussed by adds an
growth
through
other
language acquisition." that language has all the characteristics and namic systems: It is dynamic complex over time both synchronically with different and
can be dimensions cognitive shown. This interrelatedness article that by Firth the social and Wagner dimension
out that dimension by pointing important in a DST the social and the individual approach combined who and their to the argued is a reaction (1998), in second
diachronically;
it is complex (syntacti subsystems that interact; lexical, cal, phonological, textual) is unpre and sometimes it develops nonlinearly to initial con it is sensitive dictable and chaotic; feedback-sensitive, self-organizing, in and adaptive; and there are attractors develop ment. It follows that SLD also is a dynamic process. to the discrepancy Larsen-Freeman among points ditions, open, taken
(SLA) has become acquisition the dimension. by cognitive looked Freeman (this issue) developmental noted path with
set the static rules of grammar (and by researchers over in language The courses). complex the in the many factors that shape ity is visible such as interlanguages, LI the and of between type overlap in and type of input and interaction, of learners' motivation, and age, aptitude, in what is evident
there was a long 1997 article the publication of Herdina andjess that in 2002. Few researchers ner's book during to have followed Larsen-Freeman seemed period is also a out, theory building and, as she points takes and therefore social process, which dynamic that after her until silence time and Furthermore, difficult. of Her the publication Mul dina and Jessner's book, A Dynamic Model of in Psycholinguis tilingualism: Perspectives of Change In this book, tics (2002). of DST the basic notions and to SLD are their application together brought for in which various multilingualism of multilingual factors interact in the emergence awareness ism. In the model, metalinguistic plays an of the the mechanics role, important although in a model impact of metalinguistic remain poorly her various contribution awareness understood. to this in dynamic In her 2006 issue, Jessner in more de further here. Asjust mentioned, and SLD tory of DST the next milestone was in the his of conversation the availability partners. new ideas take time to settle, and shift in perspective such as this is
respect
to the applica
a fundamental
context,
of SLD nonlinearity to as in behavior" referred "U-shaped tense of regular and of the past the acquisition system is typi interlanguage irregular verbs. The in that on the basis of input, cally self-organizing the learner that reflects a system highly idiosyncratic develops of her de the unique characteristics warns (1997) to the direct respect theories against applicability over of
DST
did
optimism
with
in cognition, shows that a DST to some leads ques pertinent perspective of the hard questions tions regarding SLD. One concerns of acquisition. Larsen the mechanisms points out that in contrast to the views on
for current
Freeman
issues solves one of the core input, which in language but also (for a different, acquisition view, see Mohanan, 1992). DST-inspired a crucial From an SLD perspective, issue iswhat constitutes As Larsen-Freeman (1997) learning. of the
on innateness, com based language acquisition can emerge from the repeated applica plexity to the tion of simple procedures that the point in the output exceeds the complexity complexity
tail; therefore, The growing for SLD Bilingualism: of a keynote by Ellis, Pienemann, emerged and the discussed cial the issue: role the social
and multilingual
and Cognitions Language publishing article with peer commentaries (in de 8c Lowie, Bot, Verspoor, 2007, with commentaries Ionin, from and Lantolf, Larsen-Freeman, van Geert). A number the interaction and between contributions Liceras, of issues
is not linear growth out, points learning simple on the basis of are backslides, stag input; there and jumps, and like the unpredictability nations, it is not clear which of avalanches, instances of input or instruction Learning lead is the to which interaction instances between of in
to this spe MLJ the cognitive and 8c Cameron), (Larsen-Freeman features (Plaza language should be handled (Verspoor and informal modeling (van
learning.
This content downloaded on Thu, 17 Jan 2013 00:05:11 AM All use subject to JSTOR Terms and Conditions
characteristics
that applied as an has emerged in the gen notion important eral literature on chaos, and dynamic complexity, so not but that far has attracted the atten systems, tion of SLD is self-organized researchers, criticality (SOC). Several aspects of SLD seem to be exam there ples of such self-organized criticality, and fore a more detailed in be order may explanation here.
phenomena that patterns of development follow by showing van Orden the power et al. (2003) law. Likewise, over time in at reac have looked changes simple to show that tion time also fit experiments they with fore can the power law pattern and that be seen as evidence of SOC they there in human
In a critical evaluation of van Orden cognition. et al. (2003), and Ratcliff Farrell, Wagenmakers, that the fact that the data show (2005) argued a power law pattern is no real proof for the ex of SOC istence in cognition since that kind of can sources from various and pattern emerge
SELF-ORGANIZED
The organize idea and
CRITICALITY
self de
to that many tend systems become critical over time was the metaphor he used to
can be
more traditional theories on by state that the account of SOC They van Orden et al. is even by underspecified, to an in the which SOC way respect ap explained does actually apply to human cognition:
of the pile will become steeper a critical it reaches level. When the next one will cause
on the grain of sand dropped or more in other avalanches; a critical state. the system has reached words, Sys tems in such critical states show behav specific is reached, pile ior: they are highly unstable, and their behavior is unpredictable. in the relation between Also, of outcomes and sand) (avalanches) put (grains one can cause a or is nonlinear: grain large start they small avalanche, and once avalanches
hazard to guess that the authors' view is that the slow driving of SOC systems (cf. adding grains of sand to the pile) corresponds to the gradual accrual of in formation and that when the threshold is passed, the
in sand pile). After (cf. avalanches system responds the response ismade, the state of the system is re laxed, and the process of information accrual starts a new stimulus is presented, (p. 114)
again when
cause more and Models may larger avalanches. on these to a based have been principles applied wide variety of processes, such as forest fires, epi and neural networks. Self demics, earthquakes, organizing neural networks that are
nition and
et al. conclude that the ideas put Wagenmakers van Orden are et forward al. and by interesting but that a specific model in which challenging, to is applied SOC of human cog specific aspects has to state changes of systems. For language learn restructuring be various could Krashen's ing, examples given: "din in the head," and Spada's (1983) Lightbown The (1999) and Bahrick's (1984) retention. yet to be developed. idea of SOC is related
in a state of
to be able to to appear criticality quickly adapt new situations. Bak's ideas on SOC have been very influential in different but he fields of research, and his followers also met with fierce resistance to their ideas (see e.g., Jensen, is hard to 1998). SOC it is the result of many prove because interacting over time. variables working One of the characteristics of systems exhibit is that ing SOC follows so-called that there ber of changes How changes: with a constant small effect ones? sizes a that they show specific pattern laws. Simply, this means power is a specific relation between the num and the size of these taking place are there many large avalanches addition of grains, and how many transformations of numbers and Log lead to a linear relation between fre are
described sian
refers to a phenomenon first (1983) Barber. She reports on a by Elizabeth visit to Russia where she had to use the little Rus Krashen she had to communicate unusual: with her hosts. She noticed
something
Russian have
By the third day also, the linguist inme was noticing a rising din of Russian in my head: words, sounds, intonations, phrases, all swimming about in the voices of the people I talked with_My overall command of would improved more in a single week than it in a month or two of intensive reading. (Barber,
1980, p. 30) With Krashen his infallible called which talent takes to find a label, catchy in the head" a learner
and size, the so-called Gutenberg-Richter quency Law Because law effects 1996). (Bak, power
(1983)
phenomenon,
after
This content downloaded on Thu, 17 Jan 2013 00:05:11 AM All use subject to JSTOR Terms and Conditions
KeesDeBot
a of extensive through period input pro no obvious seem to sentences reason, cessing. For as if the system has reached in the head build has gone a critical stage into productive search providing is available. at but tem. will least does not that turns receptive knowledge. evidence the Unfortunately, for this interesting idea knowledge no re idea
173
the dynamical approach is not some wholly new way of doing research that is separate from all existing re in cognitive science and hopes to search paradigms displace approach tual map them. Rather, to see that there is a dynamical is to see a way of redrawing one's concep of cognitive science in accordance with the various forms of existing
However,
that input suggests lead to gradual changes rather it builds up to a critical and massive It suggests or
is intriguing and in a second language stage in the system, that leads of the sys itself system that into
to a sudden
restructuring
develop
that the language in such a way self-organize is turned automatically knowledge knowledge. version of and
In terms, taking a DST practical the study of language development a shift of focus, as Larsen-Freeman out point shift away in their contribution from
on to
to this issue.
It is a
"Breakthrough": because proficiency tion seems to fall stantial point eign research to the fact language and
this restructuring is what to as the refer sudden is no students jump informa real in
studies of large-scale comparative effects and from the type of experi single-factor mental reductionism that has dominated parts of in the last decades. Two examples may and Meng (2006) clarify this point. Macaro on a on the effect of report study code-switching on words learners of English learning by Chinese as a the Macaro Language. Foreign Interestingly, was and Meng part of a symposium presentation at the 2006 AAAL on the role of the conference field to help LI LI in language teaching, sentations listed variables use: 1. Attitudes 2. Level toward LI, L2, and and several an that have of the pre on impact our
sub
inquiries of a for
recall
Lightbown
is not With the sys nonuse, through tem can fall apart the feeling of having again. Also, a to is not related directly breakthrough language it is not the case that knowing 3,500 proficiency; structures and being able to perform words basic will the per hypothesis and to hypothesis have been suggested explain of retention skills (Bahrick, long-term language is The claim that there is a 1984; Neisser, 1984). in at which the system is re development in such a way that becomes knowledge and almost immune rather than sensi state phase in common all change. What is that the state or fac of the law to this experience. the threshold Similarly, lead
through indicate
mastore
3. Age 4. Type of learners 5. Amount of code-switching 6. Teachers' beliefs 7. Individual 8. Definition setting Macaro switching, and Meng and they focused compared
of proficiency of learners
of learners
and use
of L1/L2
point
on one three
is not related to a phase specific parameter tor, but results from the self-organization system. The idea that have variation interesting variation. and of not research is whether follows
a power
pattern may our about thinking on large deviations One in of the lines future to a the near leads
the translation of a (i.e., giving code-switching a definition word in the LI), in the L2 giving the translation in the LI, and giving a def plus in the L2 only. inition In a carefully designed no differences quasi-experimental among study, the conditions disappointing were found, which was of course for the researchers, but not really In an FL classroom many variables surprising: play a role in the of and the use acquisition vocabulary, factors
developed
target language of smaller and pattern larger that show a power law pattern sugges changes tive of an SOC state for the system in language input development.
is to be to many other likely secondary that play a role in the process. In a way, itwould have been if such a surprising single fac tor had in success. differences explained learning This is not a critique of the Macaro and Meng iswell and carried out care study, which designed fully. The fallacy of null effect found focusing single there are clearly many setting in which factors. tially relevant on a merely the supports in factor explaining poten
of the LI
This content downloaded on Thu, 17 Jan 2013 00:05:11 AM All use subject to JSTOR Terms and Conditions
of the study), is from a DST problematic reasons. for but different The perspective prob to be that the focus was lem with this study seemed scription too much learners. was one dents on the group rather than on individual In the the present author study, in which more of the researchers, than 12,000 stu tested on language proficiency an extensive tests. In addition, ques so contact language through media, and school charac status, attitudes, were from a re on their
the process. reason to worry about because large variations not need a do they specific explanation?this is how systems work universally. Larsen-Freeman and Cameron, however, as well as van Geert issue), et al. argue for detailed and Verspoor can account of what for the variation a specific out focus on phases with
were
pointed
2004; Wagenmakers can only prove through titative research iments empirical research.
(Jordan, paradigm
itself by
outcomes
One is that perspective came out as a very "country" important explana to link other itwas not possible tory variable, while to of the very large set used variables explain what The other was that only a country actually meant. of variance could (17-25%) very limited amount to Macaro be explained. So in contrast and Meng in this study, there were many variables (2006), but the impact of most of these variables could not be shown. likely that individual tracks are influenced the target learners' by factors devel like It is very
2002; Rueckle, (Holden, 2002), exper can be set up to show the relevance of for core aspects DST of applied such linguistics, as but there is a certain risk bilingual processing, too to strict set jumping quickly experimental on at "Concentration tings: experimental design the formative stage of model building brings ob in vious detail to the fore. That makes all into p. 237). it difficult the other to sub intuition, metaphor apply tle understandings that go (Holland, building" that while we gued our 1998, and
factors, interact
different ports
for individual
and attitudes, language, of how but the patterns over time are likely to be learners. This finding sup and Cameron of de
should
understanding we may not have reached a stage in development, can be set which meaningful up to experiments in of a DST show the relevance compe approach to current models and tition with or in addition accounts.
intriguing earlier
of variation
as part of these patterns; not need a specific explanation, more than small changes. As Bak of
may be inher are that large changes do such, large changes or at any rate not
to the exam it, referring (1996) puts narra extinction of species, the "specific ple but the each tives may explain large catastrophe, not to be confused with periodicity, sug regularity, same mechanisms work on all scales, that the gests
that the DST model mainly, but it could be argued can be to various other aspects. Language applied is a prime of a dy example policy development In two issues of the MLJ, various namic process. aspects of language policy have been and U.S. discussed 2005, extensively: heritage vol. 89, edited by Heidi languages Byrnes) (Winter,
the extinctions every day, to taking place the Cambrian the largest one, causing explosion, of all species" of up to 95 percent the extinction a way, this clashes with one of the ma (p. 18). In in perspective that Larsen-Freeman jor changes et al. (this and Cameron (this issue) and Verspoor from a focus on studies issue) suggest: microgenetic the role of context and environment in which is taken into account and in which narratives
vol. 2007, (Summer, policy language As the arti Blake and Claire Kramsch). by Robert cles in these issues show, political systems display of dynamic all the characteristics systems, with time, and sensitivity the emer
of seem from the implication of complexity No Child rules and procedures (e.g., ingly simple of the development Left Behind policy). Analysis
This content downloaded on Thu, 17 Jan 2013 00:05:11 AM All use subject to JSTOR Terms and Conditions
KeesDeBot
framework the DST using a of the and present picture more than accurate that is process developmental as to a linear analysis presented policy approach for the analysis of framework in Cooper's (1989) activities. planning language of language policies be elucidating discourse may applied and Stelma Cameron (2004) analysis. to the use of metaphors, ideas from DST applied that many other aspects showed and their analyses could and be Knox show and fruitfully (2005) all studied have from Another area in which DST be
175
The statis of the system" trinsic dynamics (p. 57). are very advanced, in these approaches tics used as a source at variation but the idea of looking to a is central of information perspec dynamic on time tive, not only in developmental processes scales ation
may
is
on short-term vari years, but also spanning in standard techniques language processing is So there and naming. like word recognition a need involved to at least a grasp of develop in this line of development. the
clearly statistics
this lan of
THE CONTRIBUTIONS
The from wide contributors different to of
IN THIS ISSUE
this special the world issue and come from
classrooms
for the analysis of processes systems, dynamic DST may provide in the classroom, of interaction at the individ a new framework in which learning ual and group levels for LI can be connected. can be Accord along by that ingly, language the lines set out van Geert. define instruction studied
Instruction
it is im that play
a parts back and paradigmatic range of theoretical an interest in the ap they share is grounds. What in the study of lan DST of principles plication sense. The in broadest the guage development one can be in two groups, contributions arranged on issues (van methodological focusing largely 8c Cameron; Larsen-Freeman Geert; Verspoor on theoret more et al.) and the other focusing ical issues (Ellis, Plaza-Pust, Jessner). van Geert from several takes Paul episodes to ex Glass the Carroll's Lewis Looking Through plain which some of the basic in his view has behind DST, thinking the momen developed In particular, for our understand not
in a model,
ably
grown consider
the multi
tum it deserves.
neuro-imaging techniques success to show with varying applied sub and neural between the relation processing strates (see de Bot, 2008, for a critical review). A somewhat raises new and maybe DST perspective in detail by van As discussed unsettling questions. lingual have been
in systems, there is no Van Geert explains of a dy the steps to be taken for the construction on a model of namic system for the L2. His focus is and growth ing of change rival theory at the moment. or While acknowledging growth development. of systems, van Geert, interconnectedness in contrast with what Larsen-Freeman what Cameron cation arrive the some and
that we need argues propose, simplifi in order to and some form of reductionism models of development of the constructs over that
at manageable
it is a simplification of The aim of the models that might takes scales explain on vari place tend to interact. it is useful and to see how
the dynamics. Development ous time scales, and these Therefore, look according on at variation at one levels. all
steps
for a move
Holden
variation on other LI
level has an impact on variation In the last section of his contribu some details about a model for
pronunciation additive that is shrouded by unsystematic, a of noise. contrast, analysis variability By cerned with identifying systematic changes ability; it assumes that variability is informative,
to empirical data. that is related acquisition Cameron and Lynne Diane Larsen-Freeman of a change focus on the specific requirements to research method from a traditional approach a to one in applied by inspired linguistics ology perspective. Their main points are that the
tion, he provides
DST
This content downloaded on Thu, 17 Jan 2013 00:05:11 AM All use subject to JSTOR Terms and Conditions
our
the behaviors
open per change. on UG as a part of the system spective language that is not completely but interacts encapsulated with other aspects of language and the social en vironment Ulrike in which Jessner, one it is used. of the of "early adopters" of that
independent interpreted Larsen-Freeman and Cameron methodological of the largely search a number principles
that of
many re about in ap
DST
cusses was
generally
and
In her
contribution,
dense
discussed with
tant role for metalinguistic of languages, multiple velopment as she admits, the implementation awareness tion of metalinguistic
is
computer modeling. Wander and Marijn Lowie, Verspoor, Marjolijn van on earlier research variation compare Dijk to variation based in SLD with a new approach as a source on DST of in that takes variation formation rather than as noise. us about and process are indicators argue of interaction a that variation informs argument the dynamics that differences The is of in
still underdeveloped. She supports her arguments on third with data from a recent study language of in Tyrol who show interesting learners patterns interaction The overview between of and within articles in this special the role DST systems. language a rich issue provide and related models Research lines: will de
the developmental ranges of variation opment. natural These outcome and in SLD,
authors
of steps of devel is the that variation of variables in de there may be "free so issue among
development
the basis
velopment variation"
that therefore
Using
ciolinguists a case
as an
development. the development of writ et al. show how Verspoor can be used to describe
language and variation will be studied on different opment of lan and decline time scales, from the growth over the life span, to variation guage proficiency
on varia focusing experimental of of variables, and modeling interactions use and Devel language development.
to and language learning. According perception, affects leads to change, Ellis's cycle, usage change affects learning, learning perception perception, he and so on. In his contribution, affects usage, links individual on lar processes the old idea and diachronic and change the societal variation level, between change and with simi reintroducing synchronic language
of variation. patterns a rich Nick Ellis presents of the emer picture a in learners of SLD from adult gence dynamic of use, language cycle language language change,
from weeks in developmental processes ranging on the level of the millisec to years, to variation as a tasks. Applied ond in experimental linguistics
is concerned with the development field basically of systems, be they individual learners, classes, eth The DST nic groups, or policy makers. perspective allows us to look at those systems in a structured tools way, with many in other fields waiting and instruments employed to be developed in ours.
of a connection language
ACKNOWLEDGMENTS
The author is indebted to Ludmila Isurin, Wander and two ML] reviewers for Verspoor, on earlier versions of this contribution.
development. In her contribution, for a the chaos link between turbulent have mirror (UG). mirror not sal Grammar
a DST
Plaza-Pust argues and Univer approach of the metaphor She uses to describe and the order systems. As she indicates, in entering very keen and their unpredictable argues for a medi
Carolina
NOTES
1 We are grateful role in the process. to Heidi Byrnes for her mediating
(p. 250).
Plaza-Pust
This content downloaded on Thu, 17 Jan 2013 00:05:11 AM All use subject to JSTOR Terms and Conditions
KeesDeBot
2The stead of the more term Second Language traditional is used in Development term Second Language growth and decline?that are both aspects Holden,
177
of response J. (2002). Fractal characteristics time variability. Ecological Psychology, 14, 53-86. Holland, J. (1998). Emergence: From chaos to order. Read ing,MA: Perseus Books. Jensen, H. (1998). Cambridge University Press. Jessner, U. (2006). Linguistic awareness in multilinguals: English as a third language. Edinburgh, UK: Edin Self-organized criticality. Cambridge:
of
REFERENCES
Bahrick, H. (1984). Fifty years of second language at research. for programmatic trition: Implications Modern Language Journal, 68, 105-118. Bak, P. (1996). How nature works: The
burgh University Press. Jessner, U. (2008). The role ofmetalinguistic knowledge in L2 and L3 development: A dynamic systems theory perspective. Modern Language Journal, 92, 270-283. Jordan, G. (2004). acquisition. Amsterdam/Philadelphia: Theory construction in second language John Ben in the head, input, and device. Foreign Language
science of self
organized criticality.New York: Copernicus. and applied Barber, E. (1980). Language acquisition ADFL 26-32. 12, Bulletin, linguistics. to cognitive sci Beer, R. (2000). Dynamical approaches
jamins. S. (1983). The din Krashen, the language acquisition Annals, 16, 41-44.
ence. Trends in Cognitive Sciences, 4, 91-99. skills Bonnet, G. (Ed.). (2004). The assessment ofpupils countries 2002. Paris: in English in eight European European Network of policy makers for the evalu systems. Browman, C, 8cGoldstein, L. (1990). Gestural specifica tion using dynamically defined articulatory struc ation of education
D. (1997). Chaos/complexity science Larsen-Freeman, and second language acquisition. Applied Linguis tics, 18, 141-165. Larsen-Freeman, language D. (2002). Language and acquisition use from a theory per chaos/complexity (Ed.), Language acquisition spective. In C. Kramsch London: and language socialization (pp. 33-46). D., 8c Cameron, language
clusters in Cameron, L., 8c Stelma, J. (2004). Metaphor discourse. Journal ofApplied Linguistics, 1, 7-36. Carroll, L. (1895). Through the looking glass. London: Macmillan. Cooper, D. (1999). Linguistic attractors: The cognitive dy namics of language acquisition and change. Amster Cooper, John Benjamins. dam/Philadelphia: R. (1989). Language planning and social change. Cambridge: Cambridge University Press. (2008). The imaging of what in the mul tilingual mind? Second Language Research, 24{A), 111-133.
tures. Journal ofPhonetics, 18, 299-320. Burns, A, 8c Knox, J. (2005). Classrooms as complex sys tems:An emergent research agenda. Paper presented at AILA World Congress, Madison, WI.
Continuum. Larsen-Freeman,
L. (2008). Research from a development Modern complex theory perspective. Language Journal, 92, 200-213. Lewin, R. (1999). Complexity: Life at the edge of chaos. methodology on
de Bot, K.
nal ofAtmospheric Sciences, 20, 130-141. Macaro, E., & Meng, Q. (2006). The effectof teacher use of LI on Chinese EFL learners' vocabulary acquisition. at the 2006 AAAL conference, Paper presented Montreal. B. (1982). The fractal geometry ofnature. New Mandelbrot, York: W. H. Freeman. Meiss,J. uary (2008) About Sci.nonlinear FAQ. Accessed Jan 12, 2008 from http://amath.colorado.edu/ of in complexity C. Ferguson, (Eds.), Phono
Chicago: The University of Chicago Press. P., 8c Spada, N. (1999). How Lightbown, languages are learned. Oxford: Oxford University Press. flow.Jour Lorenz, E. (1963). Deterministic non-periodic
de Bot, K., Verspoor, M., & Lowie, W. (2007). A Dynamic to second language ac Systems Theory approach and Cognition, quisition. Bilingualism: Language 10, 7-21. Ellis, N. (2008). The dynamics of second language emer gence: Cycles of language use, language change, and language acquisition. Modern Language Jour nal, 92, 232-249. as a dynamical Elman, J. (1995). Language system. In as motion: R. Port 8c T. van Gelder (Eds.), Mind Explorations of the dynamics of cognition (pp. 195 225). Cambridge, MA: The MIT Press. lexi Elman, J. (2004). An alternative view of the mental Firth, A, con. Trends in Cognitive Sciences, 8, 301-306. 8cWagner, J. (1998). SLA property: No tres
research, implications logical development: Models, Timonium, MD: York Press. (pp. 635-662). Neisser, U. (1984). Interpreting Harry Bahrick's discov confers immunity against forgetting? ery: What Journal ofExperimental Psychology: General, 113, 32 35. C. (2008). Universal Systems Theory: to development Grammar and Dynamic up a turbulent mirror Holding in grammars. Modern Language
Plaza-Pust,
passing! Modern Language Journal, 82, 91-94. P., 8c Jessner, U. Herdina, (2002). A dynamic model of multilingualism. Perspectives of change in psycholin guistics. Clevedon, UK: Multilingual Matters.
Journal, 92, 250-269. I. (1997). The end of certainty: Time, chaos and Prigogine, the new laws of nature. New York: Free Press.
This content downloaded on Thu, 17 Jan 2013 00:05:11 AM All use subject to JSTOR Terms and Conditions
Developmental
Rueckle, J. (2002). The dynamics of visual word recog nition. Ecological Psychology, 14, 5-19. Schueler, G., 8c Schueler, B. (1997). The chaos of Jung's 20, 2008 from http:// psyche. Retrieved March www.schuelers.com/ChaosPsyche/par_l_3htm. and dy Thelen, E., & Bates, E. (2003). Connectionism namic systems: Are they really different? Develop mental Science, 6, 378-391. Thelen, Smith, L. B. (1994). A dynamic systems approach to the development of cognition and action. Cambridge, MA: MIT Press. E., &
van Geert, P. (1995). Growth dynamics in development. In R. Port 8c T. van Gelder (Eds.), Mind as motion: in the dynamics of cognition (pp. 313 337). Cambridge, MA: Bradford Book. van Geert, P. (1998). A dynamic systems model of basic mechanisms: Piaget, Vygotsky and developmental Review, 5, 634-677. beyond. Psychological Explorations
M., Lowie, W., & van Dijk, M. (2008). Variabil from a dynamic systems per ity in L2 development Modern spective. Language Journal, 92, 214-231. E., Farrell, S., & Ratcliff, R. (2005). Hu Wagenmakers, man cognition and a pile of sand: A discussion on serial correlations ity. Journal 108-116.
van Geert, P. (2008). The Dynamic Systems approach in the study of LI and L2 acquisition: An introduc tion. Modern Language Journal, 92, 179-199. van Gelder, T. (1998). The dynamical hypothesis in cog and Brain Sciences, 21, nitive science. Behavioral 615-628.
M. (1992). Complexity: The emerging science at Waldrop, the edge of order and chaos. New York: Simon 8c Schuster. S. (2002). A new kind of science. Champaign, Wolfram, IL: Wolfram Media.
Survey of Doctoral
The annual and
Degrees
Soon To Be Available
Online
and linguistics, in foreign cultures, literatures, survey of doctoral languages, granted degrees in the fall issue of the in the United education in foreign States, which appears usually language to its readers as a service since 1926. For the last 30 years, has been by the journal published MLJ, Dr. Benseler has this extensive and has been P. Benseler Dr. David listing. However, editing compiling decided the new would to hand editorial over office our this feature and will sincerest starting be made the listing from now on will be compiled this year, and by on our new Web We site at http://mlj.miis.edu/. available to the MLJ to Dr. Benseler for his long service and the appreciation
like to extend
profession.
This content downloaded on Thu, 17 Jan 2013 00:05:11 AM All use subject to JSTOR Terms and Conditions